In a sensor the size used in a cameraphone, this number of pixels is quite meaningless. The noise levels present mean that under anything less than bright sunlight the actual information content is vastly less than the pixel count would suggest. This incessant increase in pixel count is simply driven by marketing demands, and has no relationship to the physics of the system or the actual quality of the result.

The camera will need to use quite agressive noise reduction algorithms (i.e. smoothing) and will then apply simialrly agressive edge enhancment algorithms to make the picture look sharp again. The result is a wholly unsatisfactory image. At the end of the day there is nothing that can be done about the physics of the system; very small pixels only receive a small number of photons, and the signal to noise can never be anything better than poor. A camera that used a 2 megapixel sensor of the same physical size (and thus pixels with four times the area, and hence roughly four times better signal to noise) would probably produce visually better results. Ironically results that appeared to have more detail. But marketing tends to not like such arguments, as being far to technical for the average punter who is quite proud to boast of his 8Mp camera's statistics.

When el Reg does reviews of cameraphones, it would be really useful to actually do some comparisons that have some scientific value. There is a story worthy of The Register's iconoclastic nature here, debunking the stupid pixel stats, rather than parroting them.